Did the universe have a beginning?

Audrey Mithani, Alexander Vilenkin

Institute of Cosmology, Department of Physics and Astronomy,
Tufts University, Medford, MA 02155, USA
Email : vilenkin@cosmos.phy.tufts.edu
Abstract
We discuss three candidate scenarios which seem to allow the possibil-
ity that the universe could have existed forever with no initial singularity:
eternal ination, cyclic evolution, and the emergent universe. The rst
two of these scenarios are geodesically incomplete to the past, and thus
cannot describe a universe without a beginning. The third, although it
is stable with respect to classical perturbations, can collapse quantum
mechanically, and therefore cannot have an eternal past.
1 Introduction
One of the most basic questions in cosmology is whether the universe had a
beginning or has simply existed forever. It was addressed in the singularity
theorems of Penrose and Hawking [1], with the conclusion that the initial sin-
gularity is not avoidable. These theorems rely on the strong energy condition
and on certain assumptions about the global structure of spacetime.
There are, however, three popular scenarios which circumvent these theo-
rems: eternal ination, a cyclic universe, and an emergent universe which
exists for eternity as a static seed before expanding. Here we shall argue that
none of these scenarios can actually be past-eternal.
Ination violates the strong energy condition, so the singularity theorems
of Penrose and Hawking do not apply. Indeed, quantum uctuations during
ination violate even the weak energy condition, so that singularity theorems
assuming only the weak energy condition [2] do not apply either. A more general
incompleteness theorem was proved recently [3] that does not rely on energy
conditions or Einsteins equations. Instead, it states simply that past geodesics
are incomplete provided that the expansion rate averaged along the geodesic is
positive: H
av
> 0. This is a much weaker condition, and should certainly apply
to the past of any inating region of spacetime. Therefore, although ination
may be eternal in the future, it cannot be extended indenitely to the past.
Another possibility could be a universe which cycles through an innite series
of big bang followed by expansion, contraction into a crunch that transitions
into the next big bang [4]. A potential problem with such a cyclic universe
1
a
r
X
i
v
:
1
2
0
4
.
4
6
5
8
v
1


[
h
e
p
-
t
h
]


2
0

A
p
r

2
0
1
2
is that the entropy must continue to increase through each cycle, leading to
a thermal death of the universe. This can be avoided if the volume of the
universe increases through each cycle as well, allowing the ratio S/V to remain
nite [5]. But if the volume continues to increase over each cycle, H
av
> 0,
meaning that the universe is past-incomplete.
We now turn to the emergent universe scenario, which will be our main focus
in this paper.
2 Emergent universe scenario
In the emergent universe model, the universe is closed and static in the asymp-
totic past (recent work includes [6, 7, 8, 9, 10]; for early work on oscillating
models see [11]). Then H
av
= 0 and the incompleteness theorem [3] does not
apply. This universe can be thought of as a cosmic egg that exists forever until
it breaks open to produce an expanding universe. In order for the model to be
successful, two key features are necessary. First, the universe should be stable,
so that quantum uctuations will not push it to expansion or contraction. In
addition, it should contain some mechanism to exit the stationary regime and
begin ination. One possible mechanism involves a massless scalar eld in a
potential V () which is at as but increases towards positive values
of . In the stationary regime the eld rolls from at a constant speed,

= const, but as it reaches the non-at region of the potential, ination begins
[12].
Graham et al. [10] recently proposed a simple emergent model featuring a
closed universe (k = +1) with a negative cosmological constant ( < 0) and
a matter source which obeys P = w, where 1 < w < 1/3. Graham et al.
point out that the matter source should not be a perfect uid, since this would
lead to instability from short-wavelength perturbations [10]. One such material
that fullls this requirement is a network of domain walls, which has w = 2/3.
Then the energy density is
(a) = +
0
a
1
(1)
and the Friedmann equation for the scale factor a has solutions of the form of
a simple harmonic oscillator:
a =
1
(
_

2
1 cos(t)), (2)
where
=
_
8
3
G|| (3)
and
=

2G
2
0
3||
. (4)
In the special case where = 1, the universe is static. Although this model is
stable with respect to classical perturbations, we will see that there is a quantum
instability [13, 14].
2
2.1 Quantum mechanical collapse
We consider the quantum theory for this system in the minisuperspace where
the wave function of the universe depends only on the scale factor a. In the
classical theory, the Hamiltonian is given by
H =
G
3a
_
p
2
a
+U(a)
_
, (5)
where
p
a
=
3
2G
a a (6)
is the momentum conjugate to a and the potential U(a) is given by
U(a) =
_
3
2G
_
2
a
2
_
1
8G
3
a
2
(a)
_
. (7)
With the Hamiltonian constraint H = 0, enforcing zero total energy of the
universe, we recover the oscillating universe solutions discussed in [10].
We quantize the theory by letting the momentum become the dierential
operator p
a
i
d
da
and replacing the Hamiltonian constraint with the Wheeler-
DeWitt equation [15]
H = 0. (8)
From the Hamiltonian in Eq. (5), the WDW equation becomes
_

d
2
da
2
+U(a)
_
(a) = 0, (9)
with the potential from Eqs. (1) and (7). Note that in quantum theory the form
of the potential (see Fig. 1) is no longer that of a harmonic oscillator. Instead,

U()

Figure 1: The potential U(a) with turning points a

+
and a

3
there is an oscillating region between the classical turning points a
+
and a

,
which are given by
a

=
1
_

_

2
1
_
, (10)
and the universe may tunnel through the classically forbidden region from a

to a = 0. The semiclassical tunneling probability as the universe bounces at a

can be determined from

1
P e
2S
WKB
(11)
where the tunneling action is
S
WKB
=
_
a
0
_
U(a)da =
9M
4
P
16||
_

2
2
+

4
_

2
1
_
ln
_
1
+ 1
_

1
3
_
. (12)
For a static universe, = 1 and a

= a
+
=
1
,
S
WKB
=
3M
4
P
32||
. (13)
Since the tunneling probability is nonzero, the simple harmonic universe cannot
last forever.
2.2 Solving the WDW equation
First let us examine the well-known quantum harmonic oscillator. In that case,
the wave function is a solution to the Schrodinger equation
1
2
_

d
2
dx
2
+
2
x
2
_
(x) = E(x). (14)
After imposing the boundary conditions () 0, the solutions represent a
discrete set of eigenfunctions, each having energy eigenvalue E
n
=
_
n +
1
2
_
.
However, in the case of the simple harmonic universe the wave function is a
solution to the WDW equation (9), which has a xed energy eigenvalue E = 0
from the Hamiltonian constraint. From the form of the potential in Fig. 1, it
seems that we must choose () 0, so that the wave function is bounded at
a . We are then not free to impose any additional condition at a = 0, or the
system will be overdetermined. The wave function in the under-barrier region
0 < a < a

is generally a superposition of growing and decaying solutions, and

we can expect that the solution that grows towards a = 0 will dominate (unless
the parameters of the model are ne-tuned; see [14] for more details).
A numerical solution to the WDW equation is illustrated in Fig. 2. It ex-
hibits an oscillatory behavior between the classical turning points and grows
by magnitude towards a = 0. This indicates a nonzero probability of collapse.
Similar behavior is found for the case of = 1, corresponding to a classically
static universe.
1
Semiclassical tunneling in oscillating universe models has been studied in the early work
by Dabrowski and Larsen [13].
4
a
U,

a
Figure 2: Solution of the WDW equation with ||/M
4
P
= .028 and = 1.3
(dashed line). The WDW potential is also shown (solid line).
One can consider a more general class of models including strings, domain
walls, dust, radiation, etc.,
(a) = +
C
1
a
+
C
2
a
2
+
C
2
a
3
+
C
4
a
4
+. . . . (15)
For positive values of C
n
, the eect of this is that the potential develops another
classically allowed region at small a. So the tunneling will now be to that other
region, but the qualitative conclusion about the quantum instability remains
unchanged. Altering this conclusion would require rather drastic measures.
For example, one could add a matter component
n
(a) = C
n
/a
n
with n 6
and C
n
< 0. Then the height of the barrier becomes innite at a 0 and
the tunneling action is divergent. Note, however, that such a negative-energy
matter component is likely to introduce quantum instabilities of its own.
3 Did the universe have a beginning?
At this point, it seems that the answer to this question is probably yes.
2
Here we
have addressed three scenarios which seemed to oer a way to avoid a beginning,
and have found that none of them can actually be eternal in the past. Both
eternal ination and cyclic universe scenarios have H
av
> 0, which means that
they must be past-geodesically incomplete. We have also examined a simple
emergent universe model, and concluded that it cannot escape quantum collapse.
Even considering more general emergent universe models, there do not seem to
be any matter sources that admit solutions that are immune to collapse.
2
Note that we use the term beginning as being synonimous to past incompleteness.
5
References
[1] S. W. Hawking and G. F. R. Ellis, The Large Scale Structure of Spacetime
(Cambridge University Press, Cambridge, England, 1973).
[2] A. Borde and A. Vilenkin, Eternal ination and the initial singularity,
Phys. Rev. Lett. 72, 3305 (1994) [gr-qc/9312022].
[3] A. Borde, A.H. Guth and A. Vilenkin, Inationary Spacetimes Are In-
complete in Past Directions, Phys. Rev. Lett. 90, 151301 (2003) [gr-
qc/0110012].
[4] P. J. Steinhardt and N. Turok, A cyclic model of the universe, Science 296,
1436 (2002) [hep-th/0111030].
[5] R.C. Tolman, Relativity, Thermodynamics and Cosmology. Oxford Univer-
sity Press (1934).
[6] G.F.R. Ellis and R. Maartens, The emergent universe: inationary cos-
mology with no singularity, Class. Quantum Grav. 21, 223 (2004) [gr-
qc/0211082v4].
[7] J.D. Barrow, G.F.R. Ellis, R. Maartens, and C.G. Tsagas, On the stability
of the Einstein static universe, Class. Quant. Grav. 20, L155 (2003) [gr-
qc/0302094].
[8] S. del Campo, E. Guendelman, A.B. Kaganovich, R. Herrera and P.
Labrana, Emergent universe from scale invariant two measures theory,
Phys. Lett. B 699, 211 (2011), [arXiv:1105.0651].
[9] P. Wu and H. Yu, Emergent universe from the Horava-Lifshitz gravity,
Phys. Rev. D81, 103522 (2010) [arXiv:0909.2821].
[10] P.W. Graham, B. Horn, S. Kachru, S. Rajendran, and G. Torroba, A Simple
Harmonic Universe, [arXiv:1109.0282].
[11] M. P. Dabrowski, Oscillating Friedmann cosmology, Ann. Phys. 248, 199
(1996) [gr-qc/9503017].
[12] D.J. Mulryne, R. Tavakol, J.E. Lidsey and G.F.R. Ellis, An emergent uni-
verse from a loop, Phys. Rev. D71, 123512 (2005) [astro-ph/0502589].
[13] M. P. Dabrowski and A. L. Larsen, Quantum tunneling eect in oscillating
Friedmann cosmology, Phys. Rev. D52, 3424 (1995) [gr-qc/9504025].
[14] A.T. Mithani and A. Vilenkin, Collapse of simple harmonic universe, JCAP
01 (2012) 028 [arXiv:1110.4096v4].
[15] B.S. DeWitt, Quantum Theory of Gravity. 1. The Canonical Theory, Phys.
Rev. 160, 1113 (1967).
6